NC-4153 Problem I _ _ VMPR,WFWARCode: 2.15 Am. Midl. NaL 143:64-69 _0 reprints

Home Range Use by Swamp (Sylvilagus aquaticus) in a Frequently Inundated Bottomland Forest

Department of Wildlife and FisheriPATRICKA"es, MississippiZOLLNERIState Universit); Mi_sissipfli State 39762 _'_t_a _ _- _'_'w"

Soulhern lIa,'dwoods Laboratory, United States Forest Service, Stoneville, 38776 _" _ _ rn AND

LEONARD A. BRENNAN s _ g _ I"%_ 1._- Department oflVildlil_ and Fisheries, Mississippi State University, Mississippi State 39762 _..g_ "_ _ Qa ABSTRaCr.--Home range size of six swamp rabbits ill south-central Arkansas was estilnated _ _ _ O" • :. bv radio- e emetry ti-om February 1991 through March 1992. The average home range size _ r-. tO

wasablesignificantlyto the large largernumberthanof obserpreviovusatiolynsreportedper rabbestimates.it in our Tsthuidys ,differencebut may alsisopartlybe exapulainedribut- _1. _(l i__ _"O_ O ba- by our inclusion of numerous locations of swamp rabbits during periods of deep inundation. _-_ _ _ • All of the individual rabbits tracked used different areas when the study site was flooded. _- _ _. _r_-.

Theseresultsprovidethefirstquandtafivedeseriptionoftheresponseofswamprabbitsto _ _ _ _ flooding. _ _H "

Throughout their range swamp rabbits (Sylvilagus aquaticus) are found in wet forested _ _0.t_ environments such as bottomland hardwood forests (Chapman and Feldhammer, 1981; _. -. McCollum and Holler, 1994). An important aspect of these enviromnents is the disturbance associated with frequent flooding (Harris and Gosselink, 1990). Clearly, inundation has negative consequences for swamp rabbits (e.g., increased embryo resorption [Conaway et al., 1960]) increased starvation (Svhila, 1929) and increased mortality from hunting (Mar- tinson et al., 1961). To survive in such seasonally flooded environments resident must respond to changing water levels, e.g., by moving to refugia during periods of inun- dation (Crawshaw and Quigley, 1990). The response of swamp rabbits to flooding has not been quantified. However, swamp rabbits have been observed resting on stumps surrounded by water (Ix)we, 1958; Terrel, 1972) and home range size has been shown to decrease when flooding was common (Kjolhaug and Wootf, 1988). These observations suggest that swamp rabbits restrict their movements in response to inundation, although they are also known to inhabit adjacent upland habitats (Zollner et al., 1996). It is unclear to what extent swamp rabbits use such neighboring habitat as refugia during inundation (McCollum and Holle_ 1994J but Allen (1985) suggested that these areas may be crucial for the preservation of small isolated populations. More broadly, the ecology of the swamp and the way it uses space is poorly un- derstood (Lowe. 1958: Gould. 1974: Allen. 1985L Only two published studies have investi-

Present address: USDA. Forest Service. North Central Research Station. Forestry Sciences Labora- tory. 5985 Highwa_ K. Rhinelander Wisconsin 54501 Present address: USDA. Forest Service. Pacific Northwest Research Station. Forestry Sciences Lab- oratory. 2770 Sherwood Lane-Suite 2A,Juneau. Alaska 99801 Present address: Tall Timbers Research Station. 13093 Henry Beadle Dr.. Tallahassee. Florida 32312

64 2000 ZOLLNER El" AL.: SWAMP RABP,1TS 65 gated the home range of this using radio-telemetry (Gould, 1974; Kjolhaug and Woolf, 1988) and they were conducted at the extreme southern and northern ends of the species' range, respectively. Other estimates of the size of home ranges of swamp rabbits are based on less reliable data such as trapping records and chases (Lowe, 1958; Toll et al., 1960; Terrel, 1972; Mullin, 1979). This dearth of data is surprising because the swamp rabbit is a popular game species in the southern portion of its range and a species of conservation concern in the northern portion of its range. Thus, the present study had two objectives: (1) to quantitatively assess how an individual's home range changes in response to flooding and (2) to estimate swamp rabbit's home range size in south-central Arkansas.

- METHODS

The study was conducted in Moro Bottoms Natural Area and adjacent private lands near Fordyce, Cleveland Co., Arkansas (33°47'N, 92°20'W). Moro Bottoms Natural Area contains a forest of bottomland hardwoods under the shared stewardship of the Arkansas Nature Gonservancy and the Arkansas Natural Heritage Program (Smith and Zollner, in press). The area was ca. 70 ha; 40 ha was mature bottomland hardwoods. Second-growth forests of mixed upland pines and bottomland hardwoods occurred to the north and east of Moro Bottoms Natural Area for several miles. Additional second growth bottomland hardwoods occurred to the south and west of the study site for several miles. Immediately to the south of Moro Bottoms Natural Area was a 10 ha bottomland site that was intensively harvested in July of 1989. Sweetgum (Liquidamber styraciflua) dominated the overstory in the bottom- land forests, whereas loblolly pine (Pinus taeda) dominated the overstory in the adjacent upland forests. Composition of overstory in all forests included oaks (Quercus spp.) and hickories (Carya spp,). Between 60 and 145 traps were set and checked daily between 12 January and 26 June 1991. Captured rabbits were fitted with a 350 g radio-collar and released. After release a rabbit was allowed 1 wk to adjust to its collar and recover from the stress of capture, after which observations of movements began. Four rabbits were collared and tracked for both the spring/summer (15 April-1 Oct.) and fall/winter seasons (1 Oct.-15 April), a fifth rabbit was tracked only during the spring/summer season and a sixth rabbit was tracked only during the fall/winter season. All rabbits were adult females except for a single adult male (#740) that was only tracked during the spring/summer season. Monitoring of rabbits began on 22Jan. 1991 and concluded on 5 Mar. 1992. Each rabbit's location was recorded once daily until death or transmitter failure. Diurnal observations occurred 5 d a week either between 0630 and 1200 or between 1200 and 1830 based on the flip of a coin. The other 2 d each week rabbits were located between 1830 and 0630. Starting times for noc- turnal monitoring periods were selected from a ranoom numbers table and triangulation stations (see below) were always visited in the same order. All locations used in analysis were considered temporally independent, based on the criterion that sufficient time elapsed between observations [or the rabbit to be able to move from one side of its home range to the other (White and Garrott. 1990:147). During the day we were typically able to locate each rabbit in a brush pile or thicket by walking around it with the antenna removed from the receiver. This was possible because the initial response of swamp rabbits to a threat is to remain motionless /Hamilton, 1955) and it allowed us to esumate the position of each individual within a few meters (alwavs <_5 rill. Hm,/evel_ to mimmize disturbance to the rabbits, we 'did not attempt to confirm locations with visual observations. Nocturnal positiom were determined using triangulation from a series of stations placed throughout the study area (Zollner. 1993). All nocturnal hearings were taken from stations within 250 m of the area where the animal was active. 66 THE AMERICANMIDLANDNATURALIST 143(1)

Locations were not estimated more precisely at night because of logistic difficulty and in- creased animal activity (Gould, 1974). The Universal Transverse Mercator (UTM) coordinates of the location of each swamp rabbit were used to estimate home range size and assess response to inundation. Each diurnal position was visited when the rabbit was not present and distance and bearing to a point on a transect were measured. UTM coordinates of the transects and the nocturnal triangulation stations were determined with a Trimble Pathfinder Professional (Trimble, Navigation Inc. Sunnyville, California) global positioning device. Ninety five percent fixed kernel estimates of home range size with smoothing selected by least squares cross-validation were generated using The Home Rar_ger program (Version 1.5; F. Hovey, 1999). For mul- timodal data such as ours, kernel techniques provide more accurate estimates of home range size (Seaman and Powell, 1996; Worton, 1989, 1995; Seaman et aL, 1999). Seasonal home ranges (fall-winter and spring-summer; Kjolhang and Woolf, 1988) were calculated only if >15 locations were available for an animal in a season. Annual home ranges were calculated for all animals for which seasonal home ranges were available in both seasons. Separate home ranges were also calculated for each rabbit using locations when the study area was inundated and when it was not. Locations were classified as in- undated when the depth of water at a permanent depth gauge (Fisher and Porter Co. Series 8) in the center of the study area was >39 cm. This depth corresponded to conditions during which more than 90% of Moro Bottoms Natural Area was covered by at least 20 cm of standing water (Zollner, 1993). The gauge recorded the depth of water hourly through- out the study and flooding occurred during both seasons. We used a t-test to test compare our estimates of annual home range with those reported in southern Illinois (Kjolhaug and Woolf, 1988). We used paired t-tests to compare the average fixed kernel estimates of home range size for our rabbits during inundated and dry times. We used the BLOSSOM software package (Slauson et aL, 1991) to develop Mul- tiple Response Permutation Procedures (MRPP) comparisons of the portion of home range used during inundated and dry conditions for each rabbit. MRPP statistics are a distribution- free tool for testing whether two groups come from the same or different populations (Zimmerman et aL, 1985; Biondini et al., 1988; Cade and Hoffman, 1990).

RESULTS AND DISCUSSION

Our 95% fixed kernel estimates of home range size were significandy larger It = 3.49. df = 3. P = 0.04l than the home ranges reported for southern Illinois by Kjolhaug and Woolf (19881. However. these 95% kernel estimates were within the range of values 10.7- 7.7 hal of published swamp rabbit home ranges (Kjolhaug and Wooff. 19881. Compmason between our observations and previous work are provided to place our work in the context of existing knowledge, but they should be interpreted with caution because of differences in techniques used to collect location data. techniques used to estimate home range area and sample sizes. Kjolhaug and Wooff C1988l attributed their small home ranges _o the high quality of the habitat in their study area. However, browse (Zollner. 19931 and pellet surveys tZollner et aL, 1996] suggest that habitat conditions are good for swamp rabbits in Moro Bottoms Natural Area and thus habitat quality can not explain our larger home range estimates. The patchy nature of good habitat (e.g., windthrows; Smith and Zollner, m pressl in the mature bottumland forest where our study was conducted may cause more movement and thus our larger home ranges. Additionally, our data were collected both at night and during the day. Sampling at night when nocturnal animals are more active can lead to larger home 68 THE AMERICAN MIDLAND NATURALIST 143(0 disappearing rapidly throughout the southeastern U.S. (MacDonald et al., 1979; Harris, 1984; Smith et aL, 1993). Indeed, the limited remaining habitat already occurs as frag- mented remnants in low elevation sites which are susceptible to inundation (Rudis and Birdsey, 1986). Thus, managers should consider both the landscape context and flooding history of an area when assessing whether it can support a viable population of swamp rabbits either by itself or in conjunction with adjacent _efugia.

Acknowledgenenls.--We recognize the Reaves family and K. L. Lucas for their help in the field. The encouragement and advice of J, H. McGuiness was valuable throughout the study. E.J. Heske and J. O. Whitaker, Jr. provided insightful comments on early drafts of this manuscript. This work was funded by the United States Forest Service_Southern Hardwoods Laboratory, Southern Forest Experiment Station cooperative agreement 19-90-045 with the following cooperators: Arkansas Fish and Game Com- mission, Arkansas Forestry Commission, Arkansas Natural Heritage Commission and Tl_e Nature Con- servancy. Preparation of this manuscript was supported by U. S. Department of Agriculture Cooperative State Research Service Grant no. 93-37101-8662 to W. E Smith.

LITERATURE CITED

ALLEN. A. W. 1985. Habitat suitability index model: swamp rabbit. U.S. Fish Wildlife Service Biological Report 82110.107/. 20 p. BIONDINI. M. E., E W, MIELKE,JR. ANDE. E REDENTE. 1988. Permutation techniques based on euclidean analysis spaces: a new and powerful statistical method for ecological research. Coenoses, 2:155- 174. CADE. B. S. ANDR. W. IriOFFMAN. 1990. Winter use of dougla._fir forests by blue grouse m Colorado.J WildL Manage., 54:471-479. CHAPMAN.C. H. ANn G. A. FELDHAMER. 1981. Sylivlagus aquaticus Mammalian ,5peaes. 151:1-4. CONAWAY.C. H.. T. S. BASKETTANDJ. E. TOLL. 1960. Embryo resorption in the swamp rabbit._ Wildl Manage., 24:197-202. CRAWSHAW,P. G. AND H. B. QUIGLEX'. 1990 Jaguar spacing, acdvlty and habitat use in a seasonativ flooded environment in Brazil. J, ZooL. 223:357-370. GOULD A. B. 1974. The home range and habitat preferences of swamp rabbits along a shell road in the interlnediate marsh of Southwestern . M.S. Thesis. Louisiana State University. Baton Rouge. 82 p. HAMILTON. W.J. 1955. Coprophagy in the swamp rabbit. J. , 36:803-304. HARRIS. L. D. 1984. Bottomiand hardwoods: Valuable. vanishing, vulnerable. Florida Coop. Ext. Serv. University of Florida. Gainesvifle. 18 p. HARMS. L. D. AYDJ. G. GOSSELINK. 1990. Cumulative impacts of bonomland hardwood forest conser- vauon on bydrology, water quality, and terrestrial wildlife, p. 260-320. In. J. G. Gosselink. C L. Lvndon and "i2 A. Muir (eds.) Ecological processes and cumulative impacts illustrated by bottomland hardwood wedand ecosystems, Lewis Publishers. Chelsa. Michigan. 708 p. HOLZMAN. S.. M.J CONROYANDJ. PICKERING.1992. Home range, movements and habitat use of coyotes in southcentral Georgia. _ Wildl. Manage., 56:139-t46. HOVEY. E R, 1999. The home ranger. Available at: http://nhsbig.inhs.uiuc.edu www/home-range.htmL KJOLHAUG. M. S. AND A. WOOLF. 1988. Home range of the swamp rabbit in Southern Illinois../2 Mam- maL. 69:194--197. LOWE. C. E. 1958. Ecology of the swamp rabbit in Georgia. _ MammaL. 3fl:116-127 MACDONALD. E O.. W. E. FRAYERANDJ. K. CLAUSER. 1979. Documentation. chronology, and future projections of bottomland hardwood habitat loss in the lower Mississippi Alluvial Plain. Vol. 1. HRB Singer. Inc. State College, Pennsylvania. 133 p. MARTINSON, R. K..J.W. HOLTEN AND G. K BRAKLAGE.1961. Age criteria and population dvnamics of the swamp rabbit in Missouri.f WildL Manage., 25:271-280. MCCOLLUM. R, C. AND N. R. HOLLER. 1994. Comparative use of floodplains by swamp rabbits.j_ Alabaraa Acad. Sd., 65:263-275. MULLIN. K. D. 1979. Aspects of the ecology of the swamp rabbit (S_liviagus aquaticus) in disturbed 2000 ZOLLNER ET ALL.: SWAMP RABBITS 67

TABLE1.--Average (SE)estimates of home range size (ha) of swamp rabbits

Annual Spring_Sttmmer Fall-Winter Inundated Dry (n =4) (n =5) (n =5) (n = 6) (n = 6) 95% Fixed kernel 3.1 (0.72) 4.3 (1.39) 1.0 (0.25) 4.9 (2.14) 1.2 (0.9)

range estimates (Holzman et al., 1992). Finally, our home range estimates were influenced by the movements of rabbits in response to inundation of the study area. There was no significant difference in home range size when the area was inundated (t = 1.56, df = 4, P = 0.18), ahh'o-ugh the tendency was for our estimates to increase during inundation (Table 1). Kjolhaug and Woolf (1988) described greatly restricted home ranges during flooded condidons, but performed no statistical analyses due to small sample sizes. These contradictory trends could be explained by different topographic or landscape con- texts of our study sites. Additionally, the continuous record of water depth in our study area allowed us to analyze inundation year round on a per observation basis rather than as a single seasonal event (Kjolhaug and Wooff, 1988). The MRPP tests showed that all of our rabbits used different areas when the bottoms were inundated than when the study area was not flooded (Table 2). During inundation two of the rabbits moved onto levees near the creek while the other four rabbits moved into adjacent uplands. This is the first quantitative assessment of swamp rabbit response to flooding and the best evidence that they do use adjacent uplands during flooding. It has been suggested that the preservation of small isolated populations of swamp rabbits may require the maintenance of suitable habitat to serve as a refuge from flooding (Alien, 1985: Whitaker and Abrell. 1986 b. Howeve_ pellet surveys in (McCollum and Holler. 1994) and Indiana [J. O. Whitaker, pers. comma failed to find evidence of swamp rabbit use of surrounding uplands, which may cause some to question the importance of adjacent uplands for swamp rabbits (McCollum mid Holler, 1994l In this study telemetry docu- mented that swamp rabbits moved into surrounding uplands during flooding in a study area where pellet surveys have also demonstrated swamp rabbit use of the adjacent uplands (Zollner et al.. 1996). These tracking data. along with the significantly larger home range sizes, suggest that maintenance of suitable habitat that can serve as a refuge from flooding may be importan_ at sites where prolonged heavy inundation is common. This may be particularly U'ue of the northern extent of the species' range where small patches of suitable habitat are typically embedded in an inhospitable landscape (McCullom and Holler, 1994). However. this con- clusion may apply broadly because forested bottomlands that swamp rabbits inhabit are

TABLE 2.--Muldple Response Permutation Procedure test of whether six swamp rabbits used different areas during inundated periods than during dry periods

Rabbit 740 Rabbit 600 Rabbit 460 Rabbit 440 Rabbit 380 Rabbit S40

Number of locations when inundated 17 26 27 57 46 40 Number of locations when dry 135 175 29 221 209 218 MRPP test statistic _g) 2.78 -64.89 -2.4 -10.48 -21.59 -17.49 P-value =0.03 <0.0001 =0.03 <0.0001 <0,0001 <0.0001 0 ,I 2000 ZOLLNER ET _'d..: SWAMP RABBITS 69

bottomland hardwoods and associated pinewoods in west-central Louisana. Unpubl. M,S. The- s_s. Northwestern State University. Natchitocbes. Louisiana• 39 p. RUDIS. V. A. AND R A. BIRDSEY. 1986. Forest resource trends and current conditions in the Lower Mississippi Valley. Southern Forest Experiment Station Resource Bulletin SO116. New Orleans, Louisiana: USDA Forest Service• 7 p. SEAMAN. D. E., J. J. MII.LSI'AUGH. B. J. KERNOHAN. G. C. BRUNDIGE. K. J, RAEDEKE ANDR. A. GITZEN. 1999. Effects of sample size on kernel home range esumates, d_ WildL Manage• 63:739-747. AND R. A. POWELL. 1996. An evaluation of the acuracv of kernel derxsirv estimators for home range analysis. Ecology, 77:2075-2985. SLAUSO_:.W. L. B. S. CADE ANDJ. D. PdCHARDS. 1991. BLOSSOM Statistical Software. Available at: http: www,mesc.usgs.gov, Blossoxrt.Doc/preparation,html. SMITH W. E. E B. HAMELANDR. E FO_ 1993. Mississippi alluvial valley forest conversion: implications for easlern Nortb Punerica. Proc. Southeast. Assoc. Fish Wildl. Ag., 47:46(_469. AND I_ A, ZOLLNER. lin press_ Seasonal habitat distribution of in an old-growth bot- toinland hardwood forest• In: E B. Hamel and rE Foti reds./. Bottomland hardwoods of the Mississippi .Mluvial Valley: characteristics and management of natural function. USDA Forest Service General Technical Report, Southern Research Station. Asheville North Carolina. SWHLA. R. D. 1929. Habits of Svlvilagus aquat_cus littoralis.J, iVtammal., 10:315-319. TERkEL. 22 L. 1972. The swamp rabbit (Sylvilag_ts aquaticusl in Indiana. Am. MidL Nat.. 87:283-295. TOLL, .] E.. T, S. B._SKEa-r A_D C. H. CONAWAY,1960. Home range, reproduction and foods of the swamp rabbit in Missouri. Am. MidL Nat., 63:398-412, WHITE. G, C. ANt3 R. A. GARROTr. 1990. Analysis of wildlife radio-tracking data. Academic Press Inc. New York. 383 p, WHITAKER,J. O. JR. A,'4DB. AmtELL. 1986• The swamp rabbit (Sylvilagus aquaticus) in Indiana. 1984- 1985. l_roc. Indiana Acad. Sci.. 95:563-570. WORTON. B.J. 1989. Kernel methods for estimating the utilization distribution in home-range studies. Ecolog_; 70:164-168. • 1995. Using monte carlo simulations to evaluate kernel-based home range esumators.J_ WildL Manage.. 59:794-800. ZIMMERMAN. G. M.. H. GOETZ ANDE W. MIELKEJR. 1985. Use of an improved staustical method for group comparisions to study effects of praire fire. Ecolog3; 66:606-611. ZOLLNER. E A. 1993. The ecology of the swamp rabbit in south-central Arkansas, M.S. tbesls. Mississippi State University, Mississippi State. 163 p. . W. E SMITH AID L. A. B_-_NNm'_.1996. Characteristics and adapuve mgnificance of latrines of swamp rabbits (Sylvilagus aquaticus)._ MammaL. 77:1049-1058.

SUBM[37rED 1 APmL 1999 ACCEPTEO23 JULY 1999